Merge branch 'line_break'

org_mode_samples/line_break/simple.org

@@ -0,0 +1,6 @@
+foo\\        
+bar
+
+
+lorem\\\        
+ipsum

src/compare/diff.rs

@@ -31,6 +31,7 @@ use crate::parser::Italic;
 use crate::parser::Keyword;
 use crate::parser::LatexEnvironment;
 use crate::parser::LatexFragment;
+use crate::parser::LineBreak;
 use crate::parser::Object;
 use crate::parser::OrgMacro;
 use crate::parser::Paragraph;
@@ -170,6 +171,7 @@ fn compare_object<'s>(
         Object::CitationReference(obj) => compare_citation_reference(source, emacs, obj),
         Object::InlineBabelCall(obj) => compare_inline_babel_call(source, emacs, obj),
         Object::InlineSourceBlock(obj) => compare_inline_source_block(source, emacs, obj),
+        Object::LineBreak(obj) => compare_line_break(source, emacs, obj),
     }
 }
 
@@ -1438,3 +1440,26 @@ fn compare_inline_source_block<'s>(
         children: Vec::new(),
     })
 }
+
+fn compare_line_break<'s>(
+    source: &'s str,
+    emacs: &'s Token<'s>,
+    rust: &'s LineBreak<'s>,
+) -> Result<DiffResult, Box<dyn std::error::Error>> {
+    let mut this_status = DiffStatus::Good;
+    let emacs_name = "line-break";
+    if assert_name(emacs, emacs_name).is_err() {
+        this_status = DiffStatus::Bad;
+    }
+
+    if assert_bounds(source, emacs, rust).is_err() {
+        this_status = DiffStatus::Bad;
+    }
+
+    Ok(DiffResult {
+        status: this_status,
+        name: emacs_name.to_owned(),
+        message: None,
+        children: Vec::new(),
+    })
+}

src/parser/line_break.rs

@@ -0,0 +1,59 @@
+use nom::bytes::complete::tag;
+use nom::character::complete::line_ending;
+use nom::character::complete::one_of;
+use nom::combinator::recognize;
+use nom::multi::many0;
+
+use super::Context;
+use crate::error::CustomError;
+use crate::error::MyError;
+use crate::error::Res;
+use crate::parser::util::get_consumed;
+use crate::parser::util::get_current_line_before_position;
+use crate::parser::util::get_one_before;
+use crate::parser::LineBreak;
+
+#[tracing::instrument(ret, level = "debug")]
+pub fn line_break<'r, 's>(context: Context<'r, 's>, input: &'s str) -> Res<&'s str, LineBreak<'s>> {
+    let (remaining, _) = pre(context, input)?;
+    let (remaining, _) = tag(r#"\\"#)(remaining)?;
+    let (remaining, _) = recognize(many0(one_of(" \t")))(remaining)?;
+    let (remaining, _) = line_ending(remaining)?;
+    let source = get_consumed(input, remaining);
+    Ok((remaining, LineBreak { source }))
+}
+
+#[tracing::instrument(ret, level = "debug")]
+fn pre<'r, 's>(context: Context<'r, 's>, input: &'s str) -> Res<&'s str, ()> {
+    let document_root = context.get_document_root().unwrap();
+    let preceding_character = get_one_before(document_root, input)
+        .map(|slice| slice.chars().next())
+        .flatten();
+    match preceding_character {
+        // If None, we are at the start of the file
+        None | Some('\\') => {
+            return Err(nom::Err::Error(CustomError::MyError(MyError(
+                "Not a valid pre character for line break.",
+            ))));
+        }
+        _ => {}
+    };
+    let current_line = get_current_line_before_position(document_root, input);
+    match current_line {
+        Some(line) => {
+            let is_non_empty_line = line.chars().any(|c| !c.is_whitespace());
+            if !is_non_empty_line {
+                return Err(nom::Err::Error(CustomError::MyError(MyError(
+                    "Not a valid pre line for line break.",
+                ))));
+            }
+        }
+        None => {
+            return Err(nom::Err::Error(CustomError::MyError(MyError(
+                "No preceding line for line break.",
+            ))));
+        }
+    }
+
+    Ok((input, ()))
+}

src/parser/mod.rs

@@ -25,6 +25,7 @@ mod latex_environment;
 mod latex_fragment;
 mod lesser_block;
 mod lesser_element;
+mod line_break;
 mod list;
 mod object;
 mod object_parser;
@@ -87,6 +88,7 @@ pub use object::InlineBabelCall;
 pub use object::InlineSourceBlock;
 pub use object::Italic;
 pub use object::LatexFragment;
+pub use object::LineBreak;
 pub use object::Object;
 pub use object::OrgMacro;
 pub use object::PlainLink;

src/parser/object.rs

@@ -23,6 +23,7 @@ pub enum Object<'s> {
     CitationReference(CitationReference<'s>),
     InlineBabelCall(InlineBabelCall<'s>),
     InlineSourceBlock(InlineSourceBlock<'s>),
+    LineBreak(LineBreak<'s>),
 }
 
 #[derive(Debug, PartialEq)]
@@ -149,6 +150,11 @@ pub struct InlineSourceBlock<'s> {
     pub source: &'s str,
 }
 
+#[derive(Debug, PartialEq)]
+pub struct LineBreak<'s> {
+    pub source: &'s str,
+}
+
 impl<'s> Source<'s> for Object<'s> {
     fn get_source(&'s self) -> &'s str {
         match self {
@@ -173,6 +179,7 @@ impl<'s> Source<'s> for Object<'s> {
             Object::CitationReference(obj) => obj.source,
             Object::InlineBabelCall(obj) => obj.source,
             Object::InlineSourceBlock(obj) => obj.source,
+            Object::LineBreak(obj) => obj.source,
         }
     }
 }
@@ -296,3 +303,9 @@ impl<'s> Source<'s> for InlineSourceBlock<'s> {
         self.source
     }
 }
+
+impl<'s> Source<'s> for LineBreak<'s> {
+    fn get_source(&'s self) -> &'s str {
+        self.source
+    }
+}

src/parser/object_parser.rs

@@ -15,6 +15,7 @@ use crate::parser::footnote_reference::footnote_reference;
 use crate::parser::inline_babel_call::inline_babel_call;
 use crate::parser::inline_source_block::inline_source_block;
 use crate::parser::latex_fragment::latex_fragment;
+use crate::parser::line_break::line_break;
 use crate::parser::object::Object;
 use crate::parser::org_macro::org_macro;
 use crate::parser::plain_link::plain_link;
@@ -31,6 +32,7 @@ pub fn standard_set_object<'r, 's>(
     not(|i| context.check_exit_matcher(i))(input)?;
 
     alt((
+        map(parser_with_context!(line_break)(context), Object::LineBreak),
         map(
             parser_with_context!(inline_source_block)(context),
             Object::InlineSourceBlock,
@@ -96,6 +98,7 @@ pub fn any_object_except_plain_text<'r, 's>(
 ) -> Res<&'s str, Object<'s>> {
     // Used for exit matchers so this does not check exit matcher condition.
     alt((
+        map(parser_with_context!(line_break)(context), Object::LineBreak),
         map(
             parser_with_context!(inline_source_block)(context),
             Object::InlineSourceBlock,

src/parser/token.rs

@@ -62,6 +62,7 @@ impl<'r, 's> Token<'r, 's> {
                 Object::CitationReference(_) => Box::new(std::iter::empty()), // TODO: Iterate over children
                 Object::InlineBabelCall(_) => Box::new(std::iter::empty()),
                 Object::InlineSourceBlock(_) => Box::new(std::iter::empty()),
+                Object::LineBreak(_) => Box::new(std::iter::empty()),
             },
             Token::Element(elem) => match elem {
                 Element::Paragraph(inner) => Box::new(inner.children.iter().map(Token::Object)),

src/parser/util.rs

@@ -59,6 +59,32 @@ pub fn get_one_before<'s>(document: &'s str, current_position: &'s str) -> Optio
     Some(&document[previous_character_offset..offset])
 }
 
+/// Get the line current_position is on up until current_position
+pub fn get_current_line_before_position<'s>(
+    document: &'s str,
+    current_position: &'s str,
+) -> Option<&'s str> {
+    assert!(is_slice_of(document, current_position));
+    if document.as_ptr() as usize == current_position.as_ptr() as usize {
+        return None;
+    }
+    let offset = current_position.as_ptr() as usize - document.as_ptr() as usize;
+    let mut previous_character_offset = offset;
+    loop {
+        let new_offset = document.floor_char_boundary(previous_character_offset - 1);
+        let new_line = &document[new_offset..offset];
+        let leading_char = new_line
+            .chars()
+            .next()
+            .expect("Impossible to not have at least 1 character to read.");
+        if "\r\n".contains(leading_char) || new_offset == 0 {
+            break;
+        }
+        previous_character_offset = new_offset;
+    }
+    Some(&document[previous_character_offset..offset])
+}
+
 /// Check if the child string slice is a slice of the parent string slice.
 fn is_slice_of(parent: &str, child: &str) -> bool {
     let parent_start = parent.as_ptr() as usize;

toy_language.txt

@@ -1,2 +1,3 @@
-before src_foo{ipsum} after
-src_bar[lorem]{ipsum}
+foo bar baz
+
+lorem ipsum